1 neofunctionalization of androgen receptor by gain-of-function

Article

Neofunctionalization of Androgen Receptor byGain-of-Function Mutations in Teleost Fish LineageYukiko Ogino,1 Shigehiro Kuraku,2 Hiroshi Ishibashi,3 Hitoshi Miyakawa,1,4 Eri Sumiya,1

Shinichi Miyagawa,1 Hajime Matsubara,5 Gen Yamada,6 Michael E. Baker,7 and Taisen Iguchi*,1

1Okazaki Institute for Integrative Bioscience, National Institute for Basic Biology, National Institutes of Natural Sciences, andDepartment of Basic Biology, SOKENDAI (The Graduate University for Advanced Studies), Okazaki, Japan2Phyloinformatics Unit, RIKEN Center for Life Science Technologies, Kobe, Japan3Department of Life Environmental Conservation, Faculty of Agriculture, Ehime University, Matsuyama, Japan4Center for Bioscience Research and Education, Utsunomiya University, Utsunomiya, Japan5Department of Aquatic Biology, Faculty of Bioindustry, Tokyo University of Agriculture, Abashiri, Japan6Department of Developmental Genetics, Institute of Advanced Medicine, Wakayama Medical University, Wakayama, Japan7Department of Medicine, University of California

*Corresponding author: E-mail: [email protected].

Associate editor: Willie Swanson

Abstract

Steroid hormone receptor family provides an example of evolution of diverse transcription factors through whole-genome duplication (WGD). However, little is known about how their functions have been evolved after the duplication.Teleosts present a good model to investigate an accurate evolutionary history of protein function after WGD, because ateleost-specific WGD (TSGD) resulted in a variety of duplicated genes in modern fishes. This study focused on theevolution of androgen receptor (AR) gene, as two distinct paralogs, ARa and ARb, have evolved in teleost lineageafter TSGD. ARa showed a unique intracellular localization with a higher transactivation response than that of ARb.Using site-directed mutagenesis and computational prediction of protein–ligand interactions, we identified two keysubstitutions generating a new functionality of euteleost ARa. The substitution in the hinge region contributes to theunique intracellular localization of ARa. The substitution on helices 10/11 in the ligand-binding domain possibly mod-ulates hydrogen bonds that stabilize the receptor–ligand complex leading to the higher transactivation response of ARa.These substitutions were conserved in Acanthomorpha (spiny-rayed fish) ARas, but not in an earlier branching lineageamong teleosts, Japanese eel. Insertion of these substitutions into ARs from Japanese eel recapitulates the evolutionarynovelty of euteleost ARa. These findings together indicate that the substitutions generating a new functionality of teleostARa were fixed in teleost genome after the divergence of the Elopomorpha lineage. Our findings provide a molecularexplanation for an adaptation process leading to generation of the hyperactive AR subtype after TSGD.

Key words: evolution of protein function, steroid hormone receptor, androgen, transcription factor, teleost, genomeduplication.

IntroductionThe steroid receptor (SR) family is an excellent model offunctional diversification through a series of duplications ofan ancestral SR gene, which generated six functionally differ-ent SRs, that is, androgen receptor (AR), estrogenreceptor (ER)-a and -b, progesterone receptor (PR), glucocor-ticoid receptor (GR), and mineralocorticoid receptor (MR)(Escriva et al. 1997; Thornton 2001; Bridgham et al. 2006;Paris et al. 2008; Baker 1997, 2011; Baker et al. 2013). SRduplications are thought to be associated with the 1R(round)- and 2R-whole-genome duplications (WGDs) earlyin the vertebrate lineage (Robinson-Rechavi et al. 2004). SRsregulate functions as diverse as reproduction, differentia-tion, development, metabolism, metamorphosis, or homeo-stasis (Markov et al. 2009). They function as ligand-activatedtranscription factors, thus providing a direct link betweenthe signaling molecules controlling such processes and

transcriptional responses. Understanding the genetic mecha-nisms after genome duplications in which new protein func-tions evolve leading to functional diversity of SRs is one of theimportant problems in endocrinology from the evolutionaryviewpoint.

Gene duplication may lead to the establishment of lineage-specific traits and to the development of novel biologicalfunctions (Ohno 1970, 1990; Hughes 2002; Lynch and Katju2004). The contribution of duplicated genes to theevolutionary novelties has been explained by the duplica-tion–degeneration–complementation model (Force et al.1999). This model postulates that duplicated copies wouldtake three different fates: They may be lost because of degen-erative mutations (defunctionalization), one copy acquires anovel function (neofunctionalization), or each copy adoptspart of the tasks of their parental gene (subfunctionalization).

� The Author 2015. Published by Oxford University Press on behalf of the Society for Molecular Biology and Evolution. All rights reserved. For permissions, pleasee-mail: [email protected]

228 Mol. Biol. Evol. 33(1):228–244 doi:10.1093/molbev/msv218 Advance Access publication October 27, 2015Downloaded from https://academic.oup.com/mbe/article-abstract/33/1/228/2579657by gueston 25 March 2018

The neofunctionalization model was popularized as a theoryexplaining evolutionary sources of novel protein function(Ohno 1970).

In teleosts, a 3R-WGD, teleost-specific genome duplication(TSGD), occurred (Chiu et al. 2004; Hoegg et al. 2004) approx-imately 350 Ma (Jaillon et al. 2004; Brunet et al. 2006), after thesplit of nonteleost actinopterygian lineages (namely, bichir,sturgeon, gar, and bowfin) from the teleost lineage, but beforethe divergence of Osteoglossomorpha (Chiu et al. 2004;Hoegg et al. 2004). This TSGD may have facilitated the mor-phological diversification and evolutionary radiation of tele-osts (Miya et al. 2001; Volff 2005). As a result of this TSGD,teleost genomes have retained many more duplicate genesand traces of their evolution than those of tetrapods (Satoand Nisida 2010). In fact, many teleosts have two paralogcopies, where only one ortholog is present in tetrapods(Wittbrodt et al. 1998). Thus, teleosts provide an excellentmodel system for studying the evolution of genes duplicatedthrough a WGD event, which is facilitated by the availabilityof a reliable basis of species phylogeny among diverse lineagesof teleost fishes (Inoue et al. 2003; Kikugawa et al. 2004; Volff2005; Sato and Nisida 2010).

After the TSGD event, additional copies of GRs, ARs, PRs,and ERbs compared with gene repertoire in other jawed ver-tebrates have been found in a number of teleosts (Ikeuchiet al. 1999, 2002; Sperry and Thomas 1999; Hawkins et al. 2000;Ogino et al. 2004; Wang et al. 2005; Stolte et al. 2006). Suchretention of both duplicated copies has been thought to haveconferred an advantage through neofunctionalization and/orsubfuntionalization (Prince and Pickett 2002).

The AR gene is a good analysis target in addressing evolu-tion of novel traits following genome duplication becauseduplication of the AR gene coincided with TSGD (Oginoet al. 2009). Two distinct paralogs of the AR (ARa andARb) have been identified in a number of teleosts includingNile tilapia, Japanese eel, Western mosquitofish, and Atlanticcroaker (Ikeuchi et al. 1999; Sperry and Thomas 1999; Oginoet al. 2004). Phylogenetic analyses indicate that ARa and ARbwere generated after the split of nonteleost actinopterygianlineages (bichir, sturgeon, and gar) from the future teleostlineage, but before the radiation of all extant teleosts includ-ing Osteoglossomorpha and Elopomorpha (Douard et al.2008; Ogino et al. 2009).

Diverse sex characters in sex attractive and reproductiveorgans have been shown to be regulated by steroid hormones,particularly androgens (Borg 1994; Quigley et al. 1995). Thebiological effects of androgens are mediated through the AR(Chang et al. 1988; Lubahn et al. 1988). Androgen-dependentsex differentiation has been studied in diverse teleosts includ-ing medaka, zebrafish, mosquitofish, eel, swordtail, stickle-back, tilapia, and others (Kuntz 1914; Oka 1931; Turner1941; Egami and Ishii 1956; Yamamoto and Egami 1974;Nakamura and Iwahashi 1982; Miura et al. 1991; Okada andYamashita 1994; Jakobsson et al. 1999; Rosa-Molinar andBurke 2002; Zauner et al. 2003; Ogino et al. 2004, 2014;Sone et al. 2005; Hossain et al. 2008; Offen et al. 2008).Based on molecular developmental analyses of teleost repro-ductive organs, the genes involved in the signaling pathways

of Sonic hedgehog (Shh), Bone morphogenetic protein(Bmp), and Wnt-b-catenin were identified as effector genesthat can interact with and/or function downstream of theandrogen-AR pathway (Ogino et al. 2004, 2011, 2014). ARgene duplication and its functional diversification may havecontributed to the evolutionary divergence of the secondarysex characteristics, possibly affecting the expression of sucheffector genes in teleost fishes.

We recently compared the transactivation property ofARs in vertebrates including a cartilaginous fish, Westernmosquitofish, and mouse (Ogino et al. 2009, 2011). Wefound that Western mosquitofish ARa and ARb functionsimilar to tetrapod ARs. However, ARa showed a uniqueintracellular localization and significantly higher transacti-vating response than that of mosquitofish ARb and tetra-pod AR (Ogino et al. 2009). The substitutions in ARa thatresult in these functional differences with ARb remain un-known. To better understand the molecular events thatproduced functional differences between ARa and ARb,we decided to study ARs from medaka, which is a well-defined model system. Here, we describe the substitutionsthat generate ARa-specific hypertransactivation and consti-tutive nuclear localization. In addition, we discuss the evo-lutionary history, including the timing of when thesesubstitutions in the ARa gene were fixed in the teleost lin-eage. Our results highlight a potential genomic event whichdrove evolutionary novelty of AR genes in the teleost lineage.

Results and Discussion

Teleost ARa Accumulated Novel Substitutions at aGreater Rate than That of Teleost ARbAlthough many teleost species contain two ARs, the details oftheir regulation in androgen physiology are not fully under-stood. Our previous studies of Western mosquitofish ARsrevealed that ARa has constitutive nuclear localization witha significantly higher transcriptional activity than ARb (Oginoet al. 2009, 2011). These data prompted us to investigate indetail the properties of ARa and ARb in medaka, a well-characterized model system for studies of fish physiology.Medaka has two distinct subtypes of ARs, namely ARa andARb. The deduced peptide sequences of ARa and ARb in themedaka genome are 687 and 744 amino acids long, respec-tively (fig. 1A). Alignment of the medaka ARs with human AR(hAR) illustrates that high similarity lies within the DNA-binding domain (DBD) and ligand-binding domain (LBD)(fig. 1A). The deduced amino acid sequences of the DBDand LBD of medaka AR cDNAs, along with other ARsequences of ray-finned fishes (Actinopterygii), for example,members of the sister clade to teleosts such as bichir(Polypteriformes), sturgeon (Acipenseriformes) and spottedgar (Lepisosteiformes), and teleosts including the teleost spe-cies that branched from the lineage leading to euteleost afterTSGD, silver arowana (Osteoglossiformes) and Japanese eel(Anguilliforms), were used to infer molecular phylogeny ofARs (fig.1B and supplementary fig. S1, SupplementaryMaterial online). In accordance with our previous study(Ogino et al. 2009), a phylogenetic analysis of the amino

229

Neofunctionalization of AR . doi:10.1093/molbev/msv218 MBE

Downloaded from https://academic.oup.com/mbe/article-abstract/33/1/228/2579657by gueston 25 March 2018

http://mbe.oxfordjournals.org/lookup/suppl/doi:10.1093/molbev/msv218/-/DC1



acid sequences in the DBD and LBD revealed that the geneduplication which gave rise to two different ARs (a and b)occurred in the actinopterygian lineage leading to tele-osts after the divergence of Acipenseriformes andLepisosteiformes, but before the split of Osteoglossiformes,although the confidence of this branch was not high (e.g.,bootstrap probability of up to 47 in maximum-likelihood[ML] analysis) (fig. 1B). This is compatible with the phyloge-netic timing of TSGD (Chiu et al. 2004; Hoegg et al. 2004).

Recently, the AR sequences at key lineages in vertebrateswere identified (Douard et al. 2008; Ogino et al. 2009).Cartilaginous fishes, the earliest branching group of livingjawed vertebrates, possess a functional AR (Ogino et al.2009). A comparison of the deduced amino acid sequencesof the DBD and LBD revealed that the shark AR shares highsimilarity to teleost ARbs, whereas similarity was compara-tively low to teleost ARas (fig. 1C). The higher divergence inamino acid sequences in teleost ARas, as judged by a longbranch in the molecular phylogenetic tree, indicates that,after the duplication that gave rise to ARa and ARb, thecoding sequence of ARa accumulated novel substitutionsat a greater rate than that of ARb (fig. 1B and supplementaryfig. S1A, Supplementary Material online). The evolutionaryrate difference between medaka ARa and ARb was quantifiedby Tajima’s nonparametric relative-rate test (Tajima 1993)using MEGA6 (Tamura et al. 2013). Using Spotted gar AR isused as an outgroup, the medaka ARa was shown to be 3.45times more rapidly evolving than the medaka ARb (supple-mentary fig. S1B, Supplementary Material online). Takentogether, these results suggest that the ARbs represent theancestral AR functions, and the ARas acquired new functions.

Sexual Dimorphic Expression of Medaka ARa andARbIn addition to the emergence of new functions of encodedproteins, differential gene expression is a major force in thespecialization of duplicate genes (Force et al. 1999). Expressionanalysis of medaka ARa and ARb in several adult tissues ofboth sexes (fig. 2A) revealed that both ARs are highlyexpressed in male kidney, in which the enzymes for androgenproduction are expressed (Kusakabe et al. 2002). Our quan-titative polymerase chain reaction (PCR) suggested that theARa is more highly expressed in the testis than in the ovary,whereas the ARb did not show such sexual dimorphic expres-sion in adult gonads. These differences in the expression pro-files of ARa and ARb suggest functional differences betweenthe medaka AR paralogs in gonadal development or differen-tiation. Previous studies on Atlantic croaker also indicate thatthe actions of different androgens may be mediated by dis-tinct AR subtypes with tissue-specific expression (Sperry andThomas 1999).

Different Transactivation Response and IntracellularLocalization of Medaka ARa and ARbTo analyze the functional differences of medaka ARa andARb, we examined the androgen-dependent transactivationof medaka ARs using a luciferase reporter assay (fig. 2B–D).

11-ketotestosterone (11KT), 5a-dihydrotestosoterone (DHT),testosterone (T), androstenedione (A), and 17a-methyl tes-tosterone (MT) induced significant increase of luciferaseactivity through medaka ARa and ARb. The stimulatory con-centrations of 11KT, DHT, T, A and MT were 10�9, 10�8,10�8, 10�8 and 10�9 M for ARa and 10�8, 10�8, 10�8, 10�7

and 10�9 M for ARb, respectively (fig. 2C). 11KT is the mostpotent endogenous androgen in teleost fishes (Miura et al.1991; Borg 1994). Maximum stimulation of 11KT was foundat 10�8 M for both ARs (fig. 2C). 11KT levels in the serum ofhuman chorionic gonadotropin-treated and -untreated maleJapanese eels ranged from 8� 10�10 to 2.6� 10�8 M (Miuraet al. 1991). Hence, the physiological concentration of 11KT issufficient to activate target gene expression through bothARa and ARb. Interestingly, transcriptional activation of thePRE/ARE reporter by medaka ARa was significantly higherthan that by medaka ARb (fig. 2B and C).

Tetrapod ARs are expressed in the cytoplasm and trans-located into the nucleus upon ligand binding (Simental et al.1991; Georget et al. 1997; Roy et al. 2001). Intracellular lo-calization of both medaka ARa and ARb was monitored bytransient transfection assays with DsRed-tagged AR expres-sion vectors using COS-7 cells (fig. 3B). The medaka ARb wastranslocated into the nucleus when 11KT was treated, as hasbeen observed for tetrapod ARs. In contrast, the medakaARa was constitutively located in the nucleus irrespective of11KT stimulation. Similar differences in nuclear localizationpattern were observed in Western mosquitofish ARs (Oginoet al. 2009, 2011). Our results indicate that medaka ARaacquired a new function as a hyperactive form of AR show-ing higher ligand-dependent transactivating capacity andconstitutive nuclear localization. These results support thehypothesis that AR gene duplicates derived from TSGD havefunctionally diversified in the teleost lineage through theprocesses of neofunctionalization, regarding intracellular lo-calization and transcriptional activity.

AR Subtype-Specific Intracellular Localization andTransactivation Were Regulated by the DBD andHinge Region, and LBD, Respectively

To investigate which domain generates the functional differ-ence between medaka ARa and ARb, we reciprocallyswapped corresponding domains between medaka ARaand ARb and then monitored the intracellular localizationand the transcriptional activity of the resulting chimericreceptors (fig. 3A and B).

In the chimera designated ARaba, the DBD and hingeregion of the ARa were replaced by those of the ARb. Thischimeric receptor gave a ligand-dependent nuclear localiza-tion similar to that seen with the wild-type ARb. In reversechimera designated ARbab, a DBD and hinge region of ARbwere replaced with those of the ARa. This reverse chimeraexhibited the constitutive nuclear localization identical to thewild-type ARa (fig. 3B). These results indicate that thesubstitution of the DBD and hinge region was sufficient toconvert an intracellular localization property between ARaand ARb.

230

Ogino et al. . doi:10.1093/molbev/msv218 MBE








FIG. 1. Structural comparison and phylogenetic analyses of AR genes. (A) Structural comparison of medaka ARa (GenBank accession number:AB252233) and ARb (AB252679) with hAR (NM_000044.2). The AR is composed of four different domains as defined by Krust et al. (1986): Ahypervariable N-terminal domain (NTD) that contains a constitutive activation function (AF-1), a central highly conserved DBD consisting of two zincfinger motifs, a hinge region and a COOH-terminal LBD that contains a hormone-dependent activation function (AF-2). The amino acid sequences ofthe C-terminal portion of the hinge region are divergent in their length and primary sequences among species. Both medaka ARs contain motifscharacteristic of the steroid hormone receptor family (Umesono and Evans 1989; Pfahl 1993; Takeo and Yamashita 1999; Todo et al. 1999) within theputative DBD and LBD. The numbers above each box refer to the position of amino acids in the putative DBD and LBD. Protein similarity of the

231



In the chimera ARaab, the LBD of the ARa was replacedwith that of the ARb. The ligand-induced transcriptional ac-tivity of this chimera receptor was reduced to that of the wild-type ARb (fig. 3A). In the other chimera, ARbba, an LBD ofARb was replaced with that of the ARa. The response of thischimeric receptor to 11KT was dramatically increased to begreater than the level of wild-type ARa (fig. 3A). These resultsindicate that the substitution of the LBD was sufficient toexchange the ligand-dependent response between ARa andARb.

ARa-Specific Substitution in the Hinge RegionContributes to Constitutive Nuclear Localization

To identify the mechanisms by which the ARa acquired anew function as an AR with constitutive nuclear localizationand higher ligand-dependent transactivation capacity, wecompared the amino acid sequences of medaka ARa andARb with those of ARs in other species, including shark,ray-finned fishes (Actinopterygii) such as bichir and sturgeon(Acipenseriformes), and the earlier branching teleosts aro-wana (Osteoglossiformes) and eel (Anguilliformes) as de-scribed below.

Nuclear import of the AR is mediated by the nuclearimport factor, importin-a, which recognizes and binds tonuclear localization signal (NLS) motifs (Cutress et al. 2008).The NLS is highly conserved among many nuclear receptors,such as AR, GR, MR, and PR (Cutress et al. 2008). The NLS hasbeen identified in the DBD and hinge region of hAR (Jensteret al. 1993; Zhou et al. 1994). The NLS is composed of twoclusters of basic amino acids (underlined) separated by tenamino acid residues: RKCYEAGMTLGARKLKK (Dingwall andLaskey 1991). It is conserved in teleost ARbs asKKCFEAGMTLGARKLKK, but not in teleost ARas. Aminoacid residues KRCFMSGMSLKGRRLKG were found in the cor-responding region of teleost ARas, except in Japanese eel andcavefish ARas (fig. 4A).

In the DBD and N-terminal portion of the hinge region, wefound six amino acids substituted in teleost ARas, exceptJapanese eel and cavefish ARas (fig. 4A and C). Threeamino acid replacements were present in the proximity ofthe NLS of teleost ARas (DMut 4: medaka ARa S370, HMut 1:medaka ARa K375, HMut 2 medaka ARa G381 in fig. 4A). The

nuclear export signal (NES) was identified in the hAR DBD(Black et al. 2001; Saporita et al. 2003). Three differences wereidentified on the NES of the teleost ARas DBD (medaka ARaDMut 2: medaka ARa A336, DMut 3: medaka ARa N340,H341 in fig. 4A).

To test which amino acid substitution contributes to thegeneration of ARa-specific constitutive nuclear localization,we introduced amino acid changes into the DBD and hingeregion of medaka ARb by site-directed mutagenesis (ARbDMut 1-4, ARb HMut 1, 2 in fig. 4A and C). The aminoacid substitutions ARb DMut 1–4 and ARb HMut 2 displayedno obvious effect on intracellular localization compared withwild-type ARb (fig. 4B, compare e–l, o, p with c, d). However,ARb HMut 1 in the hinge region (ARb G456K, G456 inmedaka ARb replaced with a lysine as in medaka ARa) alteredintracellular localization (fig. 4B-m, n). This mutation inducedthe constitutive nuclear localization as observed in wild-typemedaka ARa (fig. 4B, compare a with m). The reverse mutantARa HMut 1 (ARa K375G, K375 in medaka ARa replacedwith glycine as in ARb) was localized in the cytoplasm, as wellas the nucleus without the ligand (fig. 4B-q). These resultssuggest that the substitution in hinge sequences leading tothe glycine to lysine replacement is important in intracellularlocalization of the ARa. This amino acid lies immediatelyupstream of the NLS composed of basic amino acids, 628-RKLKK-632 in the hAR hinge region. The residues 620-EAGMTLGA-627 between the minor- and major-NLS clusterof hAR contact importin-a, although this interaction is weak(Cutress et al. 2008). The substitution HMut1, a nonpolaramino acid glycine to a basic amino acid lysine, in medakaARa lies in this importin-a contact region. This substitutionmay strengthen the binding of ARa to importin-a. We couldnot see significant changes in transactivation when medakaARb G456 was exchanged with lysine (ARb HMut 1) com-pared with wild-type medaka ARb (supplementary fig. S2,Supplementary Material online). This is consistent with evi-dence that a reduced binding affinity of AR for importin-ainfluences nuclear import, but has little effect on subsequenttransactivation (Cutress et al. 2008).

The mutation in the N-terminal portion of the hingeregion (HMut 1) of medaka ARb shifts it from ligand depen-dent- to independent-nuclear localization, although some dis-tribution in the cytoplasm was still observed (fig. 4B-m). A

FIG. 1. Continueddeduced amino acid sequences of each domain with hAR is shown in boxes. Protein motifs showing the similarities to those of hARs were indicated.Posterior-box (P-box) and distal-box (D-box) in the zinc fingers play roles in DNA-binding specificity and homodimerization, respectively (Umesono andEvans 1989). The NH2-terminal conserved motif interacts with the COOH terminus of the HSP70-interacting protein (CHIP) promoting AR degradation(He et al. 2004). Sumoylation sites are indicated by yellow triangles. The medaka ARs do not contain FXXLF and WXXLF sequences that mediate theNH2-terminal interaction with the LBD of the AR (He et al. 2000), and repeats of glutamine, proline, and glycine in NTD. (B) Molecular phylogenetic treebased on the amino acid sequences of actinopterygian AR genes. This tree was estimated as the ML tree, assuming JTT model. Nonteleost actinopter-ygian (bichir) AR gene was used as an outgroup. Support values at nodes are bootstrap probabilities with the JTT model and CAT+LG model in the MLanalysis, in order. A hyphen “-” indicates that the phylogenetic relationship at the particular node was not supported in the ML tree under the CAT+LGmodel. Names of the species used in these analyses and their accession numbers retrieved from GenBank and Ensembl are shown in supplementarytable S1, Supplementary Material online. (C) Percent similarity of the deduced amino acid sequences of the DBD, hinge region, and LBD of teleost ARs tobrown-banded bambooshark AR (AB213019). Alignment of the brown-banded bambooshark with other teleost ARs illustrates that the high similaritywithin the DBDs and the LBDs, sharing 79–91% identity with teleost ARa DBDs, 89% with teleost ARb DBDs, 63–69% with teleost ARa LBDs, and 68–73% with teleost ARb LBDs. The amino acid sequence of brown-banded bambooshark AR is highly similar to that of medaka ARb but much less thanthat of medaka ARa.

232








similar distribution of ARs in both nucleus and cytoplasm inthe absence of ligand stimulation was detected in the hinge-exchange chimeras ARaaba and ARbbab (supplementaryfig. S2B, Supplementary Material online). These results suggestthat there might be additional factors that influence nuclearimport of the ARa.

Key Substitution in Helices 10/11 in the ARa LBDPromotes the Higher Transactivation Response

The key residues that participate in interaction with ligandsand coactivators in hAR are conserved in both teleost ARaand ARb, as described below. Based on the three-dimen-sional (3D) structure of hAR, it was reported that the LBD ofthe AR folds into 12 helices and a ligand-binding pocket isformed by helices 3, 4, 5, 7, 11, and 12 together with the b-sheet preceding helix 6 (Matias et al. 2000; Sack et al. 2001;Gelmann 2002). In hAR, ligand binding induces folding ofhelix 12 to form a groove that binds coactivators (Gelmann2002). Thr residues composing the coactivator interface andcharge clamp for interaction with coactivator of hAR

located at helices 3, 4, 5, and 12 (Hur et al. 2004) were en-tirely conserved in both teleost ARas and ARbs (fig. 5A).Helices 4, 5, and 10 are known as primary AR contact regionsfor ligand binding. Residues that contact with steroid back-bone in hAR (Q712, M746, and R753 at the A ring; L705 andN706 at the C ring; and N706 and T878 at the D ring)(Sack et al. 2001; Gelmann 2002) were also conserved inboth teleost ARas and ARbs (fig. 5A). Helices 10/11 playa key role in NRs. The C-terminal portion of the helices10/11 forms a part of the ligand-binding cavity.Residues in the helices 10/11 of hAR, L874, F877, T878,and L881 are known to make van der Waals contact withDHT (Mongan et al. 2002). These residues were alsoconserved in both teleost ARas and ARbs (fig. 5A). Theseresults are compatible with our experimental data that bothteleost ARas and ARbs retain responses to DHT and otherandrogens.

To identify the mechanisms by which ARa was trans-formed into the hyperactive form of AR, we compared thesequences of the medaka ARa and ARb LBDs with those of

FIG. 2. Characterization of medaka ARs. (A) The relative expression of medaka ARa and ARb mRNAs in adult organs of males (blue bars) and females(red bars). The expression level was normalized to the expression of RPL7. Gene expression data are shown as the mean� SD. Different letters indicatesignificant difference in gene expression (P< 0.05) (ANOVA followed by the Tukey–Kramer test). (B, C) Ligand-dependent transactivation property ofmedaka ARs in COS-7 cells. The COS-7 cells were transfected with pCMV-medaka ARa or pCMV-medaka ARb, and pGL3 PRE/ARE tk Luc as a reporterplasmid. pRL-SV40 was used as an internal control to calculate the transfection efficiency. (B) Transfected cells were treated with the 10�9 M of variousandrogens, 11KT, DHT, T, A, MT or vehicle only as a control (indicated as C). The relative transcriptional activity of AR was shown as values normalizedby the pRL-induced activities. Vertical bars, mean� SD. Different letters indicate statistically significant difference in luciferase activity (P< 0.05)(ANOVA followed by the Tukey–Kramer test). A significant difference in AR activity based on the reporter activation was observed between ARa andARb. (C) Dose–response relationship for transcriptional activation by medaka ARa and ARb activated by DHT, 11KT, T, A, and MT. The transfectedcells were incubated with increasing concentrations of androgens (10�11–10�6 M) or vehicle only for 12 h. Vertical axis indicates fold activation, which isthe luciferase activity normalized by control of each experiment. Vertical bars represent the mean� SD. Plus signs indicate statistically significantdifference in luciferase activity compared with vehicle only (P< 0.05) (ANOVA followed by the Tukey–Kramer test). (D) EC50 and 95% confidenceintervals of EC50 of receptor activation.

233






other species ARs. We found eight conservative amino acidreplacements by those with different chemical propertieswithin the LBD in teleost ARas, except in Japanese eel andcavefish ARas (ARa LMut 1–6, ARa LMut #3, and ARa LMut#5 in fig. 5A). Based on this observation, we hypothesized thatreplacement(s) of one or more amino acids that influencedARa activity, probably occurred in the teleost lineage after thesplit of Anguilliformes. To analyze which amino acid replace-ment(s) potentially contributes to the hyperactive form ofteleost ARa, we constructed a series of medaka mutant ARasin which these amino acids were replaced with the corre-sponding amino acids from ARb by site-directed mutagenesis(fig. 5B and D).

Three substitutions, ARa LMut 1 in helix 4 (ARa D494N),ARa LMut 6 (ARa Y643F) in helices 10/11, and ARa LMut #5(ARa S663G) in helix 12, had major effects on the transcrip-tional activation of ARa. Transcriptional activities of ARaLMut 1 and ARa LMut 6 were reduced to that of wild-typeARb (fig. 5B). The transcriptional activity of reverse mutantARb LMut 6 (ARb F702Y, the F702 in medaka ARb wasreplaced with a tyrosine as in ARa) increased to a similarlevel of wild-type ARa (fig. 5C), suggesting that a single sub-stitution of F702 in medaka ARb with tyrosine (ARa Y643) issufficient to convert the ARb into the hyperactive form ofARa. The ARa LMut #5 completely lost its transactivationcapacity (fig. 5B). However, its reverse mutant ARb LMut #5(ARb G722S) did not affect transcriptional activation of ARb(fig. 5C). These results indicate that substitution of ARa S663in helix 12 is not sufficient to convert transactivation of ARbinto that of ARa, but is necessary to maintain ARa-specifictransactivation.

Upon ligand binding, helix 12 seals the ligand-bindingpocket and completes the assembly of the coactivator bind-ing groove that regulates ligand-dependent transactivation inhAR. The amino acid corresponding to ARa Y643 is F879 inhAR, which is located very close to the hAR L874, F877, T878and L881, which have van der Waals contacts with the ligandin hAR (fig. 5A). It may be possible for the substitution from aphenylalanine to tyrosine in medaka ARa in this region tomodify the ligand–protein interaction, because the aminoacid sequences within helices 10/11, 874-LHQFTFDL-881 ofhAR are highly conserved in primate ARs and teleostARbs. The structural framework for this substitution in thestability of ligand–AR interaction is discussed on the basis of3D structures of medaka ARa and ARb LBDs in the followingsection.

The mutant hAR, T878A which was originally identified inthe LNCaP metastatic prostate cancer cell line (Veldscholteet al. 1992), has a role in the discrimination of the steroid D-ring (Matias et al. 2000; Sack et al. 2001). The T878A mutationallows the AR to be activated by cortisol or other steroids andeven anti-androgens such as flutamide, thereby promotingprostate cancer cell growth (Taplin et al. 1999; Wang, Young,et al. 2000; Zhao et al. 2000). hAR H875 and D880 residues alsomutated in prostate cancer cells (Taplin et al. 1995; Duff andMcEwan 2005), although these residues have not been di-rectly implicated in steroid binding from the LBD crystalstructures (Duff and McEwan 2005). Helices 10/11 are alsoimportant for the positioning of helix 12, which is critical forcoactivator binding (Mongan et al. 2002). It seems likely,therefore, that the structural change of helices 10/11 willrender the receptor hyperactivation.

FIG. 3. Transcriptional activity and intracellular localization of the chimeric receptors in which each domain is exchanged between medaka ARa andARb. (A) Ligand-dependent transactivation of medaka chimeric ARs. The COS-7 cells were transfected with pCMV-medaka ARa, pCMV-medaka ARbor pCMV-medaka chimeric ARs, and pGL3 PRE/ARE tk Luc as a reporter plasmid. pRL-SV40 was used as an internal control to calculate the transfectionefficiency. Transfected cells were treated with the 10�8 M of 11KT or vehicle only as a control. The relative transcriptional activity of AR was shown asvalues normalized by the pRL-induced activities. Bars represent the mean� SD. Different letters indicate statistically significant difference of luciferaseactivity (P< 0.05) (ANOVA followed by the Tukey–Kramer test). (B) Intracellular localization of medaka chimeric ARs compared with the pattern ofwild-type ARa and ARb. The COS-7 cells were transiently transfected with pCMV-medaka ARa-DsRed, pCMV-medaka ARb-DsRed and series ofpCMV-medaka chimeric AR-DsRed, then treated with (+) or without (�) 11KT (10�8 M). Images represent red fluorescence for DsRed fused ARs andblue fluorescence for nuclear staining by Hoechst 33342.

234



Single Substitution in Helices 10/11 Induced theConformational Change of the Ligand-Binding Pocket,Leading to the Generation of the Hyperactive ARaSubtype

To investigate the structural basis for the different transcrip-tional properties of medaka ARa and ARb, we constructed3D models of these receptors LBDs complexed with 11KT

using a homology-modeling program of MOE-Dock basedon the crystal structure of hAR-LBD with methyltrienoloneas a template (fig. 6). In agreement with the transcriptionalproperties, the interaction energy of 11KT–ARa complexwas �33.49 kcal/mol, which was lower than that of 11KT–ARb (�22.66 kcal/mol), indicating more stable binding of11KT to ARa.

FIG. 4. Mutations in the DBD and hinge region of teleost ARas and their impact on the intracellular localization of medaka ARs. (A) Multiple alignmentsof amino acids sequences of the DBD and N-terminal portion of the hinge region in vertebrate ARs. Positions of six conservative replacements withinthose regions in teleost ARas (ARa DMut 1–4, ARa HMut 1 and 2) are indicated by black arrows. The cysteine residues for the two zinc finger motifsare shown by asterisks. P-box and D-box contained in zinc finger motifs are indicated. Note that Japanese eel ARa do not have above-mentioned sixsubstitutions. (B) Intracellular localization of medaka-mutated ARs. The COS-7 cells were transiently transfected with series of pCMV-medaka-mutatedAR-DsRed and treated with (+) or without (�) 11KT (10�8 M). In this study, six amino acids of medaka ARb indicated by arrows in panel (A) weremutated to the corresponding amino acids commonly used in teleost ARa, except Japanese eel and cavefish ARas. Note that the substitution in thehinge sequences HMut 1 leading to G456 in medaka ARb was replaced with a lysine as in medaka ARa, resulted in the localization shift to nucleuswithout ligand (m). When the K375 in medaka ARa was replaced with a glycine as in ARb, the ARa was localized in cytoplasm as well as nucleuswithout ligand (q). Images represent red fluorescence for DsRed-fused ARs and blue fluorescence for nuclear staining by Hoechst 33342. (C) Mutationsites in medaka AR DBD and hinge region designated DMut and HMut. Position of amino acids in wild type (WT) and mutants are indicated.

235



FIG. 5. Transactivation analysis of medaka ARs by inserting the subtype-specific substitutions of teleost ARs. (A) Multiple alignments of amino acidssequences of the LBD in vertebrate ARs. Positions of eight conservative amino acid replacements within the LBD in most teleost ARas (ARa LMut 1–6,ARa LMut #3 and #5) are indicated by black arrows. Residues corresponding the coactivator interface and charge clamp for interaction with coactivatorof AR are shown by yellow and blue arrowheads, respectively. Residues contacting with the A ring, C ring, and D ring of steroid backbone in hAR areindicated by magenta, green, and red arrows, respectively. Residues making van der Waals contact with ligands in helices 10/11 are marked by redasterisks. (B) Ligand-dependent transactivation capacities of medaka-mutated ARas, compared with wild-type ARa and ARb. In this experiment, eightamino acids of medaka ARa indicated by arrows in panel (A) were mutated to the corresponding amino acids used in teleost ARbs. The two

236



Several key residues in the ligand-binding pocket of thehAR-LBD, N706, R753, and T878, play a critical role in elec-trostatic interactions (e.g., hydrogen bonds) with DHT(Pereira de Jesus-Tran et al. 2006; Tan et al. 2015). In particular,R753 can establish a strong interaction with the oxygen atom(O3) of the ketone group at position C3 of androgens (Tanet al. 2015). Our analysis revealed that three residues N472,R519, and T642 in ARa (corresponding to N706, R753,and T878 in hAR) and one residue R578 in ARb (correspond-ing to R753 in hAR) would form hydrogen bonds with 11KT(fig. 6A). However, the key residues regulating the transactiva-tion capacity of medaka ARs, ARa Y643 and ARb F702, arenot directly implicated in 11KT binding (fig. 6B). To investi-gate possible structural alterations due to mutating theseresidues, we constructed a 3D model of ARa Y643F (fig.6A). Substitution of Y643 in medaka ARa with phenylalaninewould abolish hydrogen bonds to 11KT of N472 and T642.Alternatively, M547 would participate in a hydrogen bondwith 11KT (fig. 6A). The calculated interaction energy ofARa Y643F with 11KT was �18.83 kcal/mol, which is higherthan that of wild-type ARa (�33.49 kcal/mol), indicating thatARa Y643F would weaken the interaction with 11KT, com-pared with wild-type ARa. Hence, a single substitution inhelices 10/11 induces the conformation change of the li-gand-binding pocket and structural alterations, leading tothe generation of the hyperactive ARa subtype after TSGD.

Recapitulation of Medaka ARa-Specific ProteinFunction from Basal Teleost ARs

The bichir and sturgeon that split before the TSGD shouldhave one AR gene before the split between ARa and ARb. Incontrast, the arowana and eel, which represent the earlierbranching teleost groups (Inoue et al. 2003; Li et al. 2008),might have ARa and ARb genes that were derived from theirrespective ancestral genes before functional diversification.Although it is likely that the arowana genome contains anunidentified ARa or had secondarily lost ARa (Ogino et al.2009), both Japanese eel ARa and ARb were in fact expressedin the cytoplasm and translocate into the nucleus uponligand stimulation (fig. 7A) and did not show significant dif-ference in transactivation response in COS-7 cells (fig. 7B).

Finally, we tested our hypothesis that the substitutionsglycine to lysine in the hinge region and/or phenylalanineto tyrosine in the LBD in euteleost ARa cause different intra-cellular localization and higher transactivation betweenteleost ARa and ARb by using mutant Japanese eel ARs(fig. 7A–C). Mutation in the hinge sequence, glycine tolysine (Eel ARa HMut 1: Eel ARa G561K; Eel ARb HMut 1:Eel ARb G510K), leads to the ligand-independent-nuclear lo-calization of Japanese eel ARs, although some distribution in

the cytoplasm was still detected (fig. 7A) as observed inmedaka mutant ARs (HMut1) (fig. 4B-m, q). The phenylala-nine replaced with a tyrosine in helices 10/11 in the Japaneseeel ARs (Eel ARa LMut 6: Eel ARa F806Y and Eel ARb LMut 6:Eel ARb F755Y) enhanced the ligand-dependent transactiva-tion compared with wild-type Japanese eel ARs (fig. 7B), in-dicating an evolutionary importance of this replacement.Together, these results suggest that the two substitutions ofamino acids from glycine to lysine in the hinge region andfrom phenylalanine to tyrosine in helices 10/11 of the LBDrecapitulate the evolution of the function of teleost ARa fromthe earlier branching teleost ARs before functionaldiversification.

Differential Retention and Loss of Duplicated ARGenes during Teleost Diversification

Duplicated ARs have been identified in a number of teleostspecies. The phylogenetic analysis of longer branches of tele-ost ARas indicated that teleost ARa accumulated substitu-tions at a greater rate than teleost ARb. In most cases,substitutions in the encoding region render one of the dupli-cates from nonfunctional to its eventual loss (Jaillon et al.2004; Brunet et al. 2006). The additional copy persists whennatural selection favors differential protein function or regu-lation of duplicates, which could be triggered by changes afterduplication, if advantageous mutations occur in one or bothparalogs, or if degenerative mutations make the paralogsnonredundant (Ohno 1970; Ferris and Whitt 1979; Sidow1996; Force et al. 1999). In this study, we found that substi-tutions in teleost ARa genes result in an increase of ligand-dependent transactivation. Such functional innovation in theARa protein could provide the selective advantage for theretention of the duplicate AR in teleost genomes.

It has turned out that Japanese eel retains the two ARsubtypes (Ikeuchi et al. 1999) having the similar level of trans-activation response and the ligand-dependent nuclear local-ization property (fig. 7A and B). Neither AR contains thesubstitutions in both the hinge region and LBD that affecteuteleost ARa protein function. Note that our phylogeneticanalysis indicates that the duplication of Japanese eel ARgenes had not occurred by lineage-specific gene duplication(supplementary fig. S1A, Supplementary Material online).Japanese eel is a member of Elopomorpha, the sister lineageof other teleosts that appeared after the TSGD (Guo et al.2010). This suggests that the change of amino acids that affectARa protein function occurred after the split of Elopomorphafrom the lineage leading to the Acanthomorpha lineage.Interestingly, the ARa genes after the split of Elopomorphacomprised the monophyletic clade (e.g., bootstrap probabilityof 99/100 in ML analysis) (fig. 1B), suggesting the possibility

FIG. 5. Continuedsubstitutions in the ARa LBD, LMut 1 (ARa D494N), and LMut 6 (ARa Y643F) prominently reduced the ligand-dependent transactivation to that of thewild-type ARb. ARa LMut #5 (ARa S663G) completely abolished this ligand-dependent transactivation. (C) Ligand-dependent transactivation capacitiesof medaka-mutated ARb. Reverse mutant ARb LMut 6 (ARb F702Y) showed the similar transcriptional activity compared with the wild-type ARa.Transfected cells were treated with 10�8 M of 11KT or vehicle only as control. Bars in (B) and (C) represent the mean� SD. Different letters indicatestatistically significant difference in luciferase activity (P< 0.05) (ANOVA followed by the Tukey–Kramer test). (D) Mutation sites in medaka AR LBDdesignated LMut. Position of amino acids in wild type (WT) and mutants are indicated.

237





that the new functionality observed in medaka ARa mighthave occurred in these ARa genes comprising the monophy-letic clade. The eel does not show the apparent externalsexual differences (Dolan and Power 1977) compared witheuteleost fishes. Our study opens up the possibility that themorphological innovation of sex characteristics in teleostsmight coincide with the evolution of AR protein functionin teleost lineage.

Hox clusters as well as NRs have been thought as appro-priate indicators of genome duplication (Garcia-Fernandezand Holland 1994; Amores et al. 1998). The eel retains eightHox clusters which originated at the TSGD (Guo et al. 2010;Henkel et al. 2012). The preservation of the ARs before thefunctional innovation of protein function and fully populated,duplicated Hox clusters in the eel imply that the functionalevolution of duplicated genes did not occur immediately afterTSGD.

Based on the mitochondrial gene phylogeny, it has beenthought that Otocephala diverged after the split of

Elopomorpha, but before the divergence of Salmoniformes(Saitoh et al. 2003). In the Otocephala, two subtypes of ARs,categorized into the two distinct clusters including ARa andARb genes, were identified in the cavefish Astyanax mexica-nus, a member of Characiformes by searching the Ensembldatabase (fig. 1B and supplementary fig. S1, SupplementaryMaterial online). However, an ARa gene has not been iden-tified in the members of Cypriniformes (e.g., zebrafish Daniorerio, gold fish Carassius auratus, and fathead minnowPimephales promelas) (Wilson et al. 2004; Hossain et al.2008) and Siluriformes (e.g., southern catfish Silurus meridio-nalis) (Huang et al. 2011). This suggests that the ARa gene wassecondarily lost in the lineages of Cypriniformes andSiluriformes.

In Salmoniformes whose lineage diverged early in euteleostevolution (Near et al. 2012), two AR subtypes were identified(Takeo and Yamashita 1999). However, these AR subtypeswere categorized into the ARb cluster indicating that theduplication of the salmonid AR gene had occurred by

FIG. 6. In silico analysis of the interaction potentials between 11KT and the 3D model of the medaka AR LBDs. (A) Predicted binding modes obtainedfrom the docking simulation analysis of 11KT for the medaka ARa-, ARb-, and ARaY643F-LBDs. The amino acid residues R519, and N472, and T642 inthe medaka ARa positioned close to the oxygen atom of the ketone group at positions C3 and C17 of 11KT, respectively. In medaka ARb, R578 closelylocated to oxygen atom at C3 position of 11KT. In the medaka ARa Y643F, R519 and M547 instead of N472 and T642 are positioned close to the oxygenatom at C3 and C17 of 11KT, respectively. Orange arrows indicate the predicted hydrogen bonds between them. Note that the amino acid residues ARaY643 and ARb F702 are positioned at a greater distance from this oxygen atom. (B) 3D ribbon diagram of medaka ARa and ARb. The helices 10/11(residues 616–648 of the medaka ARa, residues 675–707 of the medaka ARb) are labeled with blue. The positions of the ARa Y643 and ARb F702 in thehelices 10/11 are indicated.

238






lineage-specific gene duplication in the recent salmonid tet-raploid event, estimated to have taken place 100–50 Ma(Allendorf and Thorgaard 1984). The ARa might have beenlost before this lineage-specific gene duplication. Taken

together, it is likely that the lineage-specific losses of theARa gene occurred independently in Cypriniformes,Siluriformes, and Salmoniformes (Douard et al. 2008;Hossain et al. 2008; Ogino et al. 2009; Huang et al. 2011).

The key substitution in helices 10/11 of the LBD, but not inthe hinge region, was observed in cavefish ARa, suggestingthe significance of the substitution in the LBD for retention ofARa although transactivation capacity and intracellular local-ization of cavefish ARa and ARb is still unclear.

If this hypothesis of independent losses of the ARa gene inCypriniformes, Siluriformes, and Salmoniformes is correct,two possible scenarios of evolution would be proposed toexplain the retention of the ARa gene in the teleost genomes(fig. 8).

The first scenario is that the key substitution of the LBDwas inserted into the ARa gene after the divergence ofElopomorpha, but before the split of the Otocephala lineage.The hyperactive form of ARa was not selected in the mem-bers of the Cypriniformes, Siluriformes, and Salmoniformes.Then, the ARa gene was lost independently in these lineages.The second possibility is that the substitutions in the LBDare presumably fixed in euteleost lineage after the split ofSalmoniformes. The substitution in the LBD might have inde-pendently occurred in a member of Characiformes. Therefore,the ARa gene has been retained in Characiformes andAcanthomorpha but not in Cypriniformes, Siluriformes, andSalmoniformes.

In summary, the gene duplication between ARa and ARboccurred before the radiation of all extant teleosts includingOsteoglossiformes and Elopomorpha. By tracing evolutionarychanges in protein function, we discovered two amino acidreplacements that generate the differences in protein func-tion between medaka ARa and ARb. Functional analysis ofthese substitutions illustrates that the amino acid changefrom glycine to lysine in the hinge region is necessary for ashift from ligand dependent- to constitutive-nuclear localiza-tion of ARa. The phenylalanine to tyrosine substitution inhelices 10/11 in the LBD was sufficient to promote theARa-specific higher transactivation capacity. 3D structuralmodeling revealed that alteration of this residue located inthe helices 10/11 composing part of the ligand-bindingpocket may modify the ligand–protein association. The gen-eration of the hyperactive ARa subtype would provide thechanges in amount of AR target genes expression, whichmight be adopted as the heterometry (Gilbert 2005), achange in amount of gene expression leading to phenotypicchange. Although functional analysis of these AR genes in vivois required to discuss the biological importance of the ARgene duplication in the teleost lineage, evolutionary occur-rence of two functionally distinct AR proteins might havefacilitated the phenotypic diversification of sex characteristicsin the euteleost fishes. The evolutionary processes of AR geneloss and retention in teleosts illustrate that teleosts representan excellent model system to study phenotypic effects ofchanges in gene repertoire.

FIG. 7. Introduction of the teleost ARa-specific substitutions intoJapanese eel ARa and ARb. (A) Intracellular localization of Japaneseeel wild type and mutated ARs. Both wild-type ARs showed theligand-dependent nuclear localization. Mutation of hinge region,HMut 1 was introduced into ARs (eel ARa G561K and eel ARbG510K). These mutated ARs were located into the nucleus withoutligand. (B) Ligand-dependent transactivation of Japanese eel ARs. Bothwild-type ARs showed no significant differences in the ligand-dependenttransactivation. The substitution in LBD (LMut 6, eel ARa L806Y, eelARb F755Y) prominently increased the ligand-dependent transactiva-tion, compared with those of the wild-type ARs. Transfected cells weretreated with the 10�8 M of 11KT or vehicle only as a control. Barsrepresent the mean� SD. Different letters indicate statistically signifi-cant difference in luciferase activity (P< 0.05) (ANOVA followed by theTukey–Kramer test). (C) Mutation sites in eel AR hinge region and LBDdesignated HMut 1 and LMut 6. Position of amino acids in wild type(WT) and mutants are indicated.

239



Materials and Methods

Animals

For gene expression analyses, Orange-red strain and Qurtstrain of medaka (Oryzias latipes) were kindly provided byDr Kiyoshi Naruse (National Bioresource Project, Medaka inNational Institute for Basic Biology) and Dr Hiroshi Mitani(University of Tokyo), respectively, and maintained at 26.5 �Cin a laboratory breeding colony. Japanese eel (Anguilla japon-ica) was kindly provided by Dr Tohru Kobayashi (University ofShizuoka). All procedures and protocols were approved bythe Institutional Animal Care and Use Committee at theNational Institute for Basic Biology.

RNA Isolation and Quantitative Reverse TranscriptionPCR

Total RNA was prepared from liver, kidney, gonad, brain,heart, gill, muscle, anal fin, caudal fin, and dorsal fin of bothOrange-red female and male medaka using ISOGEN (NipponGene, Tokyo, Japan). Changes in gene expression were con-firmed and quantified using the 7500 Real-time PCR system(Life Technologies, Carlsbad, CA). One microgram of totalRNA was reverse transcribed and amplified using

SuperScript III and SYBR Green master mix (LifeTechnologies), respectively. Relative RNA equivalents foreach sample were determined by normalizing to the expres-sion of RPL7 (Zhang and Hu 2007). Gene expression levels ineach tissue were compared. Error bars represent standarddeviation (SD). Statistical analysis was performed by analysisof variance (ANOVA) followed by the Tukey–Kramer posthoc test using Excel 2011 (Microsoft Corp., Redmond, WA)with an add-in software Statcel 3 (Yanai 2011). Primer setsused for the quantitative PCR analyses were as follows, ARa-rF1: CGGCGACCGAACTTTCAG, ARa-rR1: TGCGGTCGGACAGGTAGACT, ARb-rF1: GGATGCCCAGGACACCCTAT, ARb-rR1: AGCCACTCACCGACCTCACT. Amplification of RPL7was performed using RPL-7-rF1: CGCCAGATCTTCAACGGTGTAT, RPL-7-rR1: AGGCTCAGCAATCCTCAGCAT (Zhangand Hu 2007) in all experiments as a control.

Phylogenetic Analysis

Phylogenetic analysis of AR genes was performed on aminoacid sequences of the DBD and LBD. The amino acidsequences were aligned by MEGA6 software using defaultoptions (Tamura et al. 2013). The ML tree was constructedfrom this alignment, assuming the JTT (Jones, Taylor, and

FIG. 8. Composite phylogeny for Actinopterygii (Volff 2005) with the hypothesized scenario of teleost AR genes evolution. The evolutionary treeillustrating the TSGD event that gave rise to two different teleost ARs, ARa and ARb. TSGD occurred in the actinopterygian lineage leading to teleostsafter the divergence of Acipenseriformes (sturgeon) and Lepisosteiformes (gar), but before the split of Osteoglossiformes (arowana), which is associatedwith the TSGD (indicated by a red line) (Douard et al. 2008; Ogino et al. 2009). The two hypothesized timings of key substitution in the LBD of ARagenes were indicated by green and blue bars, and the estimated timing of secondary loss of ARa (orange circles) was designated by comparison ofprimary sequences and mutational analyses.

240



Thorton) model using MEGA6 and the CAT+LG model usingRAxML. Confidence in the phylogeny was assessed by boot-strap resampling of the data with 1,000 replications(Felsenstein 1985). Names of the species used in these anal-yses and their accession numbers retrieved from GenBankand Ensembl are shown in supplementary table S1 andfigure S1, Supplementary Material online.

Construction of Plasmid Vectors

Medaka AR cDNAs (ARa: AB252233; ARb: AB252679) werecloned into CMV expression vectors, pCS2+MT or pDsRedmonomer-N1 (Life Technologies), producing pCMV-medakaARa, pCMV-medaka ARb, pCMV-medaka ARa-DsRed, andpCMV-medaka ARb-DsRed, as previously described (Katohet al. 2006). A reporter plasmid for AR (PGL3 PRE/ARE tk Luc)was utilized (Matsui et al. 2002). Site-directed mutagenesiswas performed using PrimeSTAR Max mutagenesis basalkit (Takara, Ohtsu, Japan). Domain-swapped variantswere generated by In-fusion Advantage PCR cloning kit(Takara). All clones were verified by sequencing. Primersutilized for the mutagenesis and domain swapping areshown in supplementary tables S2 and S3, SupplementaryMaterial online.

Gene Transfection Assay

COS-7 cells cultured in 24-multiwell plates (5.0� 104 cells/well) were transfected with 400 ng/well of PRE/ARE reporterplasmid, 0.8 ng/well of pRL-SV40 (Renilla luciferase vector) asthe internal control, and 80 ng/well of the expression vectorfor ARs, using 1.5ml/well of TransFast transfection reagent(Promega, Tokyo, Japan). After 6 h of transfection, cells wereincubated for 12 h in Dulbecco’s Modified Eagle Medium(DMEM) with 10% charcoal-treated fetal bovine serum inthe presence or absence of DHT (Wako Chemical, Tokyo,Japan, 045-26071), T (Wako Chemical, 204-08343), A(Sigma-Aldrich, Tokyo, Japan, A-9630), 11KT (Sigma-Aldrich,K-8250), and MT (Sigma-Aldrich, M-7252).

The reporter gene activities were determined by thedual-luciferase reporter assay system (Promega) with valuesnormalized by pRL-induced activities (i.e., firefly luciferase ac-tivity/Renilla luciferase activity). All experiments were re-peated no less than three times. Statistical analysis wasperformed by ANOVA followed by the Tukey–Kramer posthoc test using Excel 2011 (Microsoft Corp.) with an add-insoftware Statcel 3 (Yanai 2011). The data are presented as themeans� SD. Localization of DsRed fusion ARs was visualizedunder a fluorescence microscope 18 h after the transfection.The nuclei were stained with 2mg/ml bisbenzimide H33342(Sigma-Aldrich, B2261).

Computational Model

Construction of a homology model for the medaka ARa- andARb-LBD and the in silico analysis of the interaction poten-tials between 11KT and the 3D model of the LBD receptorwere performed using the Molecular Operation Environment(MOE) software (Chemical Computing Group Inc., Montreal,QB, Canada), as described previously (Uchida et al. 2015). The

crystal structure of the hAR-LBD bound to methyltrienolone(Protein Data Bank entry 1E3G) (Matias et al. 2000) was usedas the template to build the homology models of bothmedaka ARa- and ARb-LBD. The amino acid sequences ofmedaka ARa (AB252233) and ARb (AB252679) were alignedwith those of 1E3G to yield a readily superimposable 3Dmodel. The structure of the medaka ARa- and ARb-LBD inthe absence of the ligand was optimized by the AMBER12:EHT force field (Wang, Cieplak et al. 2000). The medaka ARa-and ARb-LBD model was prepared with the Protonate 3Dprogram by adjusting the protonation state of the runningbuffer to pH 7.0 in which the 11KT–AR interaction was mon-itored. These complex structures were then used to identifythe ligand-binding sites using MOE Alpha Site Finder. Thechemical structure of 11KT was constructed, rendered, andminimized with the MMFF94x force field in MOE. The mo-lecular volumes of the resulting ligand structures were calcu-lated with the AtomRegion program using a grid space of0.1 A. Possible docking of 11KT was searched with theASEDock program. All ASEDock algorithms were codedusing MOE Scientific Vector Language. Existing features im-plemented in MOE were completely applied to realize theASEDock functions. A total of 250 conformations were gen-erated for each chemical by LowMode Molecular Dynamics(Labute 2010). The most stable ligand-binding modes of 11KTwith the medaka ARa- and ARb-LBD were determined basedon the lowest U-total value (kcal/mol). Each docking simula-tion was evaluated as a U-dock score (kcal/mol) (U_ele[electric energy] + U_vdw [van der Waals energy] + U_solv[solvation energy] + U_strain [strain energy]). The criticalamino acids for the ligand interaction were also determinedby the MOE ligand interaction module.

Supplementary MaterialSupplementary figures S1 and S2 and tables S1–S3 are avail-able at Molecular Biology and Evolution online (http://www.mbe.oxfordjournals.org/).

Acknowledgments

The authors thank Dr Kiyoshi Naruse, Dr Minoru Tanaka, DrTohru Kobayashi, Dr Shoji Oda, Dr Tatsuya Sakamoto, DrHirotaka Sakamoto, Dr Minetaro Ogawa, Dr HiroshiSakamoto, Dr Masato Kinoshita, and Dr Yasuhiro Kamei fortheir suggestions. This study was supported by Grants-in-Aidfor Scientific Research (C) (23570085) (15K07138) to Y.O. andScientific Research (B) (24370029) (15H04396) to T.I., Y.O.,and S.M. from Ministry of Education, Culture, Sports,Science and Technology, Japan; grants from Ministry of theEnvironment, Japan to T.I. and S.M.; National Institute forBasic Biology to T.I.; and the program of the Joint Usage/Research Center for Developmental Medicine from Instituteof Molecular Embryology and Genetics, KumamotoUniversity, Japan to Y.O. This work was also supported bythe Center for the Promotion of Integrated Science (CPIS) ofThe Graduate University for Advanced Studies (SOKENDAI),Japan to Y.O.

241













http://www.mbe.oxfordjournals.org/

http://www.mbe.oxfordjournals.org/

References

Allendorf FW, Thorgaard GH. 1984. Tetraploidy and the evolution ofSalmonid fishes. In: Turner JB, editor. Evolutionary genetics of fishes.New York: Plenum Publishing Corp. p. 1–53.

Amores A, Force A, Yan YL, Joly L, Amemiya C, Fritz A, Ho RK, LangelandJ, Prince V, Wang YL, et al. 1998. Zebrafish hox clusters and verte-brate genome evolution. Science 282:1711–1714.

Baker ME. 1997. Steroid receptor phylogeny and vertebrate origins. MolCell Endocrinol. 135:101–107.

Baker ME. 2011. Origin and diversification of steroids: co-evolution ofenzymes and nuclear receptors. Mol Cell Endocrinol. 334:14–20.

Baker ME, Funder JW, Kattoula SR. 2013. Evolution of hormone selec-tivity in glucocorticoid and mineralocorticoid receptors. J SteroidBiochem Mol Biol. 137:57–70.

Black BE, Holaska JM, Rastinejad F, Paschal BM. 2001. DNA bindingdomains in diverse nuclear receptors function as nuclear exportsignals. Curr Biol. 11:1749–1758.

Borg B. 1994. Androgens in teleost fishes. Comp Biochem Physiol.109C:219–245.

Bridgham JT, Carroll SM, Thornton JW. 2006. Evolution of hor-mone-receptor complexity by molecular exploitation. Science312:97–101.

Brunet FG, Roest Crollius H, Paris M, Aury JM, Gibert P, Jaillon O, LaudetV, Robinson-Rechavi M. 2006. Gene loss and evolutionary rates fol-lowing whole-genome duplication in teleost fishes. Mol Biol Evol.23:1808–1816.

Chang CS, Kokontis J, Liao ST. 1988. Molecular cloning of human and ratcomplementary DNA encoding androgen receptors. Science240:324–326.

Chiu CH, Dewar K, Wagner GP, Takahashi K, Ruddle F, et al. 2004. BichirHoxA cluster sequence reveals surprising trends in ray-finned fishgenomic evolution. Genome Res. 14:11–17.

Cutress ML, Whitaker HC, Mills IG, Stewart M, Neal DE. 2008. Structuralbasis for the nuclear import of the human androgen receptor. J CellSci. 121:957–968.

Dingwall C, Laskey RA. 1991. Nuclear targeting sequences—a consensus?Trends Biochem Sci. 16:478–481.

Dolan JA, Power G. 1977. Sex ration of American eels, Anguilla rostrata,from the Matemark river system, Quebec, with remarks on prob-lems in sexual identification. J Fish Res Board Can. 34:294–299.

Douard V, Brunet F, Boussau B, Ahrens I, Vlaeminck-Guillem V,Haendler B, Laudet V, Guiguen Y. 2008. The fate of the duplicatedandrogen receptor in fishes: a late neofunctionalization event? BMCEvol Biol. 8:336.

Duff J, McEwan IJ. 2005. Mutation of histidine 874 in the androgenreceptor ligand-binding domain leads to promiscuous ligand acti-vation and altered p160 coactivator interactions. Mol Endocrinol.19:2943–2954.

Egami N, Ishii S. 1956. Sexual differences in the shape of some bones inthe fish, Oryzias latipes. J Fac Sci., Tokyo Univ., IV7:563–571.

Escriva H, Safi R, Hanni C, Langlois MC, Saumitou-Laprade P, Stehelin D,Capron A, Pierce R, Laudet V. 1997. Ligand binding was acquiredduring evolution of nuclear receptors. Proc Natl Acad Sci U S A.94:6803–6808.

Felsenstein J. 1985. Confidence limits on phylogenies: an approach usingthe bootstrap. Evolution 39:783–791.

Ferris SD, Whitt GS. 1979. Evolution of the differential regulation ofduplicate genes after polyploidization. J Mol Evol. 12:267–317.

Force A, Lynch M, Pickett FB, Amores A, Yan YL, Postlethwait J. 1999.Preservation of duplicate genes by complementary, degenerativemutations. Genetics 151:1531–1545.

Gelmann EP. 2002. Molecular biology of the androgen receptor. J ClinOncol. 20:3001–3015.

Garcia-Fernandez J, Holland PW. 1994. Archetypal organization of theamphioxus Hox gene cluster. Nature 370:563–566.

Georget V, Lobaccaro JM, Terouanne B, Mangeat P, Nicolas JC, Sultan C.1997. Trafficking of the androgen receptor in living cells with fused

green fluorescent protein-androgen receptor. Mol Cell Endocrinol.129:17–26.

Gilbert SF. 2005. Putting evo-devo into focus. An interview with Scott F.Gilbert. Interview by Alexander T. Mikhailov. Int J Dev Biol. 49:9–16.

Guo B, Gan X, He S. 2010. Hox genes of the Japanese eel Anguillajaponica and Hox cluster evolution in teleosts. J Exp Zool B MolDev Evol. 314:135–147.

Hawkins MB, Thornton JW, Crews D, Skipper JK, Dotte A, Thomas P.2000. Identification of a third distinct estrogen receptor and reclas-sification of estrogen receptors in teleosts. Proc Natl Acad Sci U S A.97:10751–10756.

He B, Bai S, Hnat AT, Kalman RI, Minges JT, Patterson C, Wilson EM.2004. An androgen receptor NH2-terminal conserved motif interactswith the COOH terminus of the Hsp70-interacting protein (CHIP). JBiol Chem. 279:30643–30653.

He B, Kemppainen JA, Wilson EM. 2000. FXXLF and WXXLF sequencesmediate the NH2-terminal interaction with the ligand bindingdomain of the androgen receptor. J Biol Chem. 275:22986–22994.

Henkel CV, Burgerhout E, de Wijze DL, Dirks RP, Minegishi Y, Jansen HJ,Spaink HP, Dufour S, Weltzien FA, Tsukamoto K, et al. 2012.Primitive duplicate Hox clusters in the European eel’s genome.PLoS One 7:e32231.

Hoegg S, Brinkmann H, Taylor JS, Meyer A. 2004. Phylogenetic timing ofthe fish-specific genome duplication correlates with the diversifica-tion of teleost fish. J Mol Evol. 59:190–203.

Hossain MS, Larsson A, Scherbak N, Olsson PE, Orban L. 2008. Zebrafishandrogen receptor: isolation, molecular, and biochemical character-ization. Biol Reprod. 78:361–369.

Huang BF, Sun YL, Wu FR, Liu ZH, Wang ZJ, Luo LF, Zhang YG, WangDS. 2011. Isolation, sequence analysis, and characterization of an-drogen receptor in Southern catfish, Silurus meridionalis. Fish PhysiolBiochem. 37:593–601.

Hughes AL. 2002. Adaptive evolution after gene duplication. TrendsGenet. 18:433–434.

Hur E, Pfaff SJ, Payne ES, Gron H, Buehrer BM, Fletterick RJ. 2004.Recognition and accommodation at the androgen receptor coacti-vator binding interface. PLoS Biol. 2:E274.

Ikeuchi T, Todo T, Kobayashi T, Nagahama Y. 1999. cDNA cloning of anovel androgen receptor subtype. J Biol Chem. 274:25205–25209.

Ikeuchi T, Todo T, Kobayashi T, Nagahama Y. 2002. A novel progestogenreceptor subtype in the Japanese eel, Anguilla japonica. FEBS Lett.510:77–82.

Inoue JG, Miya M, Tsukamoto K, Nishida M. 2003. Basal actinopterygianrelationships: a mitogenomic perspective on the phylogeny of the“ancient fish.” Mol Phylogenet Evol. 26:110–120.

Jaillon O, Aury JM, Brunet F, Petit JL, Stange-Thomann N, Mauceli E,Bouneau L, Fischer C, Ozouf-Costaz C, Bernot A. 2004. Genomeduplication in the teleost fish Tetraodon nigroviridis reveals theearly vertebrate proto-karyotype. Nature 431:946–957.

Jakobsson S, Borg B, Haux C, Hyllner SJ. 1999. An 11-ketotestosteroneinduced kidney-secreted protein: the nest building glue from malethree-spined stickleback, Gasterosteus aculeatus. Fish PhysiolBiochem. 20:79–85.

Jenster G, Trapman J, Brinkmann AO. 1993. Nuclear import of thehuman androgen receptor. Biochem J. 293(Pt 3):761–768.

Katoh H, Ogino Y, Yamada G. 2006. Cloning and expression analysis ofandrogen receptor gene in chicken embryogenesis. FEBS Lett.580:1607–1615.

Kikugawa K, Katoh K, Kuraku S, Sakurai H, Ishida O, Iwabe N, Miyata T.2004. Basal jawed vertebrate phylogeny inferred from multiple nu-clear DNA-coded genes. BMC Biol. 2:3.

Krust A, Green S, Argos P, Kumar V, Walter P, Bornert JM, Chambon P.1986. The chicken oestrogen receptor sequence: homology with v-erbA and the human oestrogen and glucocorticoid receptors. EMBOJ. 5:891–897.

Kuntz A. 1914. Notes on the habits, morphology of the reproductiveorgans, and embryology of the viviparous fish Gambusia affinis. BullUS Bur Fish. 33:177–190.

242



Kusakabe M, Kobayashi T, Todo T, Mark Lokman P, Nagahama Y,Young G. 2002. Molecular cloning and expression during sperma-togenesis of a cDNA encoding testicular 11b-hydroxylase (P45011b) in rainbow trout (Oncorhynchus mykiss). Mol Reprod Dev.62:456–469.

Labute P. 2010. LowModeMD—implicit low-mode velocity filtering ap-plied to conformational search of macrocycles and protein loops.J Chem Inf Model. 50:792–800.

Li C, Lu G, Orti G. 2008. Optimal data partitioning and a test case for ray-finned fishes (Actinopterygii) based on ten nuclear loci. Syst Biol.57:519–539.

Lubahn DB, Joseph DR, Sar M, Tan J, Higgs HN, Larson RE, French FS,Wilson EM. 1988. The human androgen receptor: complementarydeoxyribonucleic acid cloning, sequence analysis and gene expres-sion in prostate. Mol Endocrinol. 2:1265–1275.

Lynch M, Katju V. 2004. The altered evolutionary trajectories of geneduplicates. Trends Genet. 20:544–549.

Markov GV, Tavares R, Dauphin-Villemant C, Demeneix BA, Baker ME,Laudet V. 2009. Independent elaboration of steroid hormonesignaling pathways in metazoans. Proc Natl Acad Sci U S A.106:11913–11918.

Matias PM, Donner P, Coelho R, Thomaz M, Peixoto C, Macedo S, OttoN, Joschko S, Scholz P, Wegg A. 2000. Structural evidence for ligandspecificity in the binding domain of the human androgen receptor.Implications for pathogenic gene mutations. J Biol Chem. 275:26164–26171.

Matsui D, Sakari M, Sato T, Murayama A, Takada I, Kim M, Takeyama K,Kato S. 2002. Transcriptional regulation of the mouse steroid 5a-reductase type II gene by progesterone in brain. Nucleic Acids Res.30:1387–1393.

Miura T, Yamauchi K, Takahashi H, Nagahama Y. 1991. Hormonal in-duction of all stages of spermatogenesis in vitro in the male Japaneseeel (Anguilla japonica). Proc Natl Acad Sci U S A. 88:5774–5778.

Miya M, Kawaguchi A, Nishida M. 2001. Mitogenomic exploration ofhigher teleostean phylogenies: a case study for moderate-scale evo-lutionary genomics with 38 newly determined complete mitochon-drial DNA sequences. Mol Biol Evol. 18:1993–2009.

Mongan NP, Jaaskelainen J, Green K, Schwabe JW, Shimura N, DattaniM, Hughes IA. 2002. Two de novo mutations in the AR gene causethe complete androgen insensitivity syndrome in a pair of mono-zygotic twins. J Clin Endocrinol Metab. 87:1057–1061.

Nakamura M, Iwahashi M. 1982. Studies on the practical masculiniza-tion in Tilapia nilotica by the oral administration of androgen. BullJpn Soc Sci Fish. 486:763–769.

Near TJ, Eytan RI, Dornburg A, Kuhn KL, Moore JA, Davis MP,Wainwright PC, Friedman M, Smith WL. 2012. Resolution of ray-finned fish phylogeny and timing of diversification. Proc Natl AcadSci U S A. 109:13698–13703.

Offen N, Blum N, Meyer A, Begemann G. 2008. Fgfr1 signalling in thedevelopment of a sexually selected trait in vertebrates, the sword ofswordtail fish. BMC Dev Biol. 8:98.

Ogino Y, Hirakawa I, Inohaya K, Sumiya E, Miyagawa S, Denslow N,Yamada G, Tatarazako N, Iguchi T. 2014. Bmp7 and Lef1 are thedownstream effectors of androgen signaling in androgen-inducedsex characteristics development in medaka. Endocrinology 155:449–462.

Ogino Y, Katoh H, Kuraku S, Yamada G. 2009. Evolutionary history andfunctional characterization of androgen receptor genes in jawedvertebrates. Endocrinology 150:5415–5427.

Ogino Y, Katoh H, Yamada G. 2004. Androgen dependent developmentof a modified anal fin, gonopodium, as a model to understand themechanism of secondary sexual character expression in vertebrates.FEBS Lett. 575:119–126.

Ogino Y, Miyagawa S, Katoh H, Prins GS, Iguchi T, Yamada G. 2011.Essential functions of androgen signaling emerged through the de-velopmental analysis of vertebrate sex characteristics. Evol Dev.13:315–325.

Ohno S. 1970. Evolution of gene duplication. New York: Springer Verlag.Ohta T. 1990. How gene families evolve. Theor Popul Biol. 37:213–219.

Oka TB. 1931. On the processes on the fin-rays of the male of Oryziaslatipes and other sex characters of this fish. J Fac Sci Imp Univ TokyoSec. 2:209–218.

Okada YK, Yamashita H. 1944. Experimental investigation of the man-ifestation of secondary sexual characters in fish, using the medaka,Oryzias latipes (Temminck and Schlegel) as material. J Fac Sci., TokyoUniv., IV6 6:383–437.

Paris M, Pettersson K, Schubert M, Bertrand S, Pongratz I, Escriva H,Laudet V. 2008. An amphioxus orthologue of the estrogen receptorthat does not bind estradiol: insights into estrogen receptor evolu-tion. BMC Evol Biol. 8:219.

Pereira de Jesus-Tran K, Cote PL, Cantin L, Blanchet J, Labrie F, Breton R.2006. Comparison of crystal structures of human androgen receptorligand-binding domain complexed with various agonists revealsmolecular determinants responsible for binding affinity. ProteinSci. 15:987–999.

Pfahl M. 1993. Nuclear receptor/AP-1 interaction. Endocr Rev. 14:651–658.

Prince VE, Pickett FB. 2002. Splitting pairs: the diverging fates of dupli-cated genes. Nat Rev Genet. 3:827–837.

Quigley CA, De Bellis A, Marschke KB, El-Awady MK, Wilson EM, FrenchFS. 1995. Androgen receptor defects: historical, clinical, and molec-ular perspectives. Endocr Rev. 16:271–321.

Robinson-Rechavi M, Boussau B, Laudet V. 2004. Phylogenetic datingand characterization of gene duplications in vertebrates: the carti-laginous fish reference. Mol Biol Evol. 21:580–586.

Rosa-Molinar E, Burke AC. 2002. Starting from fins: parallelism in theevolution of limbs and genitalia: the fin-to-genitalia transition. EvolDev. 4:124–126.

Roy AK, Tyagi RK, Song CS, Lavrovsky Y, Ahn SC, Oh TS, Chatterjee B.2001. Androgen receptor: structural domains and functional dy-namics after ligand-receptor interaction. Ann N Y Acad Sci.949:44–57.

Sack JS, Kish KF, Wang C, Attar RM, Kiefer SE, An Y, Wu GY, Scheffler JE,Salvati ME, Krystek SR, Jr. 2001. Crystallographic structures of theligand-binding domains of the androgen receptor and its T877Amutant complexed with the natural agonist dihydrotestosterone.Proc Natl Acad Sci U S A. 98:4904–4909.

Saitoh K, Miya M, Inoue JG, Ishiguro NB, Nishida M. 2003. Mitochondrialgenomics of ostariophysan fishes: perspectives on phylogeny andbiogeography. J Mol Evol. 56:464–472.

Saporita AJ, Zhang Q, Navai N, Dincer Z, Hahn J, Cai X, Wang Z. 2003.Identification and characterization of a ligand-regulated nu-clear export signal in androgen receptor. J Biol Chem. 278:41998–42005.

Sato Y, Nisida M. 2010. Teleost fish with specific genome duplication asunique models of vertebrate evolution. Environ Biol Fish. 88:169–188.

Sidow A. 1996. Gen(om)e duplications in the evolution of early verte-brates. Curr Opin Genet Dev. 6:715–722.

Simental JA, Sar M, Lane MV, French FS, Wilson EM. 1991.Transcriptional activation and nuclear targeting signals of thehuman androgen receptor. J Biol Chem. 266:510–518.

Sone K, Hinago M, Itamoto M, Katsu Y, Watanabe H, Urushitani H, TooiO, Guillette LJ, Jr., Iguchi T. 2005. Effects of an androgenic growthpromoter 17b-trenbolone on masculinization of mosquitofish(Gambusia affinis affinis). Gen Comp Endocrinol. 143:151–160.

Sperry TS, Thomas P. 1999. Characterization of two nuclear androgenreceptors in Atlantic croaker: comparison of their biochemical prop-erties and binding specificities. Endocrinology 140:1602–1611.

Stolte EH, van Kemenade BM, Savelkoul HF, Flik G. 2006. Evolution ofglucocorticoid receptors with different glucocorticoid sensitivity.J Endocrinol. 190:17–28.

Tajima F. 1993. Simple methods for testing the molecular evolutionaryclock hypothesis. Genetics 135:599–607.

Takeo J, Yamashita S. 1999. Two distinct isoforms of cDNA encodingrainbow trout androgen receptors. J Biol Chem. 274:5674–5680.

Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. 2013. MEGA6:Molecular Evolutionary Genetics Analysis version 6.0. Mol Biol Evol.30:2725–2729.

243



Tan MH, Li J, Xu HE, Melcher K, Yong EL. 2015. Androgen receptor:structure, role in prostate cancer and drug discovery. ActaPharmacol Sin. 36:3–23.

Taplin ME, Bubley GJ, Ko YJ, Small EJ, Upton M, Rajeshkumar B, Balk SP.1999. Selection for androgen receptor mutations in prostate cancerstreated with androgen antagonist. Cancer Res. 59:2511–2515.

Taplin ME, Bubley GJ, Shuster TD, Frantz ME, Spooner AE, Ogata GK,Keer HN, Balk SP. 1995. Mutation of the androgen-receptor gene inmetastatic androgen-independent prostate cancer. N Engl J Med.332:1393–1398.

Thornton JW. 2001. Evolution of vertebrate steroid receptors from anancestral estrogen receptor by ligand exploitation and serial genomeexpansions. Proc Natl Acad Sci U S A. 98:5671–5676.

Todo T, Ikeuchi T, Kobayashi T, Nagahama Y. 1999. Fish androgenreceptor: cDNA cloning, steroid activation of transcription in trans-fected mammalian cells, and tissue mRNA levels. Biochem BiophysRes Commun. 254:378–383.

Turner CL. 1941. Morphognesis of the gonopodium in Gambusia affinisaffinis. J Morphol. 69:161–185.

Uchida M, Ishibashi H, Yamamoto R, Koyanagi A, Kusano T, TominagaN, Ishibashi Y, Arizono K. 2015. Endocrine-disrupting potentials ofequine estrogens equilin, equilenin, and their metabolites, in themedaka Oryzias latipes: in silico and DNA microarray studies. JAppl Toxicol. 35:1040–1048.

Umesono K, Evans RM. 1989. Determinants of target gene specificity forsteroid/thyroid hormone receptors. Cell 57:1139–1146.

Veldscholte J, Berrevoets CA, Ris-Stalpers C, Kuiper GG, Jenster G,Trapman J, Brinkmann AO, Mulder E. 1992. The androgen receptorin LNCaP cells contains a mutation in the ligand binding domainwhich affects steroid binding characteristics and response to anti-androgens. J Steroid Biochem Mol Biol. 41:665–669.

Volff JN. 2005. Genome evolution and biodiversity in teleost fish.Heredity 94:280–294.

Wang C, Young WJ, Chang C. 2000. Isolation and characterization of theandrogen receptor mutants with divergent transcriptional activityin response to hydroxyflutamide. Endocrine 12:69–76.

Wang DS, Senthilkumaran B, Sudhakumari CC, Sakai F, Matsuda M,Kobayashi T, Yoshikuni M, Nagahama Y. 2005. Molecular cloning,

gene expression and characterization of the third estrogen receptorof the Nile tilapia, Oreochromis niloticus. Fish Physiol Biochem.31:255–266.

Wang J, Cieplak P, Kollman P. 2000. How well does a restrained elec-trostatic potential (RESP) model perform in calculating conforma-tional energies of organic and biological molecules? Comput Chem.21:1049–1074.

Wilson VS, Cardon MC, Thornton J, Korte JJ, Ankley GT, Welch J, GrayLE, Jr., Hartig PC. 2004. Cloning and in vitro expression and charac-terization of the androgen receptor and isolation of estrogen recep-tor a from the fathead Minnow (Pimephales promelas). Environ SciTechnol. 38:6314–6321.

Wittbrodt J, Meyer A, Schartl M. 1998. More genes in fish? Bioessays20:511–515

Yamamoto M, Egami N. 1974. Fine structure of the surface of the analfin and the processes on its fin rays of male Oryzias latipes. Copeia1974:262–265.

Yanai H. 2011. Statcel—the useful add-in software forms on Excel. 3.Tokyo (Japan): OMS.

Zauner H, Begemann G, Mari-Beffa M, Meyer A. 2003. Differentialregulation of msx genes in the development of the gonopo-dium, an intromittent organ, and of the “sword,” a sexuallyselected trait of swordtail fishes (Xiphophorus). Evol Dev.5:466–477.

Zhang Z, Hu J. 2007. Development and validation of endogenous refer-ence genes for expression profiling of medaka (Oryzias latipes) ex-posed to endocrine disrupting chemicals by quantitative real-timeRT-PCR. Toxicol Sci. 95:356–368.

Zhao XY, Malloy PJ, Krishnan AV, Swami S, Navone NM, Peehl DM,Feldman D. 2000. Glucocorticoids can promote androgen-indepen-dent growth of prostate cancer cells through a mutated androgenreceptor. Nat Med. 6:703–706.

Zhou ZX, Sar M, Simental JA, Lane MV, Wilson EM. 1994. A ligand-dependent bipartite nuclear targeting signal in the human androgenreceptor. Requirement for the DNA-binding domain and modula-tion by NH2-terminal and carboxyl-terminal sequences. J Biol Chem.269:13115–13123.

244



1 neofunctionalization of androgen receptor by gain-of-function

Documents